This protocol is reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Protocols (PRISMA-P)  and uses an integrated knowledge translation approach throughout the conduct of the systematic review. Each component of the project includes stakeholders to ensure relevance of the project for parents, clinicians, and decision-makers. The protocol has been registered in PROSPERO (Registration # CRD42013005426), an international register of prospective systematic review protocols.
Search methods for study identification
A comprehensive search strategy was developed by an experienced information specialist and peer reviewed using the Peer Review of Electronic Strategies tool . The electronic databases MEDLINE, EMBASE, CINAHL, PsycINFO, and the Cochrane CENTRAL database of controlled trials will be searched. A search strategy developed for MEDLINE is shown in Additional file 1 and will be adapted as required for each database.
‘Grey literature’ searches will be conducted for other potentially relevant articles. We will hand search recently published issues of key journals (for example, Ear and Hearing, Journal of Deaf Studies and Deaf Education) for additional studies not yet indexed in electronic databases. Proceedings and abstracts from key conferences (Newborn Hearing Screening Conference, International Cochlear Implant Symposium in Children, and A.G. Bell Conference) as well as relevant professional websites (A.G. Bell Association for the Deaf, National Acoustics Lab, National Health Services (United Kingdom) Newborn Hearing Screening Programme) will also be searched to identify unpublished studies.
It is anticipated that this review will include primarily non-randomized studies as well as observational studies as our previous review in this field of study yielded only one randomized controlled trial . Accordingly, we will include the following study designs: randomized clinical trials, controlled clinical trials and other quasi-experimental designs that include comparator groups, prospective cohort studies, and retrospective cohort studies. Case-control, cross-sectional, case series, and case studies will be excluded.
Included studies will meet the following criteria: 1) children with permanent hearing loss of any degree of severity and of early onset (prior to age 2 years) and typically using hearing aids or cochlear implants; and 2) children identified and enrolled in early intervention programs by 2 years of age.
Studies that include only outcomes of children with developmental disabilities that involve cognitive delay in addition to hearing impairment will be excluded.
We will include studies addressing early intervention aimed at spoken language development, at least comprised of: 1) therapy involving oral language stimulation; and 2) therapy including any form of sign language or sign support (for example, Signing Exact English, ASL, Langue des signes québeçoise, sign assist, baby sign language), which may form part of an intervention program (for example, total communication, dual communication, simultaneous communication, bilingual approaches).
Relevancy of papers will be assessed on the basis of the components of the approach (that is, spoken language intervention with or without some form of sign language included), and not on the basis of the program label.
Control groups will include children receiving early intervention spoken language therapy without sign language (that is, hearing technology and oral language stimulation).
Primary outcomes will include all measures of listening and spoken language development including auditory skills (for example, speech perception tests), oral receptive and expressive vocabulary and language, speech production, and speech intelligibility. These outcomes are well supported in the recent literature as clinically relevant outcomes [13, 18, 25]. However, during the broad and focused screening stages, articles will not be excluded on the basis of outcomes.
Electrophysiologic outcomes (for example, auditory brainstem or cortical responses) will also be examined to document potential “objective” benefits of various types of intervention.
Any adverse outcomes as reported in studies will be collected.
Given our interest on the effects of visual (sign) languages on early identified children, we will only include studies published 1995 onward as previous generations of children were unlikely to receive the same standards of care related to early identification of hearing disorders and access to new hearing technologies (for example, cochlear implantation).
For feasibility, only articles written in the English and French languages will be included. Articles in other languages will be screened at the broad screening level for potential relevance and the details of any relevant citations will be included in the final report.
Once all records have been retrieved through electronic and other searching methods, they will be compiled in a Reference Manager database and checked for duplication. All remaining citations will then be exported to the Distiller Systematic Review Software (DSR), an internet-based software program [http://systematic-review.net/] for the study selection process.
The study selection involves two specific stages. Screening forms will be developed from the inclusion and exclusion criteria and calibrated among reviewers with a subset of records before each screening stage takes place.
Titles and abstracts will be assessed by one reviewer for potential relevance; a second reviewer will verify those records deemed not relevant.
Two independent reviewers will screen all potentially relevant full-text articles. Disagreements will be resolved by consensus or by consultation with a third member of the research team when needed.
Electronic study-specific data abstraction forms will be used to abstract pre-determined data for each study. Data will be collected and managed in DSR. Data abstraction items will broadly include: 1) study characteristics (author names and contact information, year, institution, country, language, publication status, source of funding); 2) study design; 3) population characteristics - for example, sample size, sex, ethnicity, etiology (including radiologic findings when available), age of hearing loss identification and intervention, severity of hearing loss, type of hearing technology, time with hearing technology, parental involvement, socio-economic status, home/intervention language, cognitive status; 4) details of intervention (including fidelity); 5) details of control or comparison groups (including fidelity); 6) risk of bias assessments; and 7) outcome definitions and data. The data abstraction form was finalized with input from the knowledge users at the first team meeting. One researcher will extract all information; a second reviewer will verify all information. Discrepant findings will be resolved through consensus or a third reviewer when required.
If information or data are missing or incomplete, we will attempt to contact the study authors twice through email over a 4-week period. We will not impute data for any outcomes.
Risk of bias assessment
The risk of bias assessment will be conducted by one researcher with full verification by a second researcher. For randomized and controlled clinical trials, the Cochrane Risk of Bias tool  will be used; domains of assessment relate to selection, performance, attrition, detection, selective reporting, and other biases. For quasi-experimental studies (interrupted time series and controlled before-after designs) we will use the Effective Practice and Organisation of Care modification of the Cochrane Risk of Bias tool. Cohort designs or case controlled studies will be assessed using the Qualitative Assessment Tool for Quantitative Studies, a tool developed by the Effective Public Health Practice Project at McMaster University to assess the quality of studies in a systematic review [35, 36]. The tool, accompanied by a reviewer's dictionary, results in an overall methodological rating of studies based on an appraisal of eight areas: selection bias, study design, confounders, blinding, data collection methods, withdrawals and dropouts, integrity of intervention and study analysis. Any disagreements in assessments will be resolved through discussion or by third party adjudication.
Study characteristics will be summarized narratively in the text and/or in tables in the report; data may be presented as frequencies and percentages, medians and interquartile ranges, or means and standard deviations, where appropriate. A narrative synthesis of the evidence will be conducted when quantitative pooling of data is not possible. Based on previous reviews [31, 37, 38] that showed great heterogeneity in research designs and wide variability in methods and types of spoken language outcomes, it is anticipated that a full meta-analysis will not be possible.
If, based on clinical similarity, quantity, and quality, meta-analysis is possible for some outcomes, we will use a random effects model to generate aggregate results. For continuous data we will compute mean or standardized mean differences with 95% confidence intervals (CIs). Any change-from-baseline data will be collected in addition to final data. Risk ratios with 95% CIs will be computed for dichotomous data. Decisions about handling and analyzing ordinal outcomes will be determined post hoc, subject to the body of evidence available. For any time-to-event data, the generic inverse variance method will be used to meta-analyze outcomes using log hazard ratios and standard errors. When required, we will convert data (for example, from standard error to standard deviation) to facilitate analyses and consistency in the presentation of study findings. Statistical heterogeneity will be evaluated using I-squared (I2) statistics; for the interpretation of I2, a rough guide of low (0-25%), moderate (25-50%), substantial (50-75%), and considerable (75-100%) heterogeneity will be used; possible reasons contributing to heterogeneity will be explored. If there is considerable heterogeneity (>75%), we will not complete a pooled analysis. If data permit, sensitivity analyses may be undertaken with respect to risk of bias (restricting to studies with low risk of bias), the fidelity of the intervention, data issues, or measurement of outcomes. If at least 10 studies are included in a meta-analysis, funnel plots will be generated to assess for publication bias and other possible reasons for asymmetry .
We will examine the following variables in subgroup analyses as effect modifiers, if feasible: 1) severity of hearing loss (mild, moderate, severe); 2) children with auditory neuropathy; 3) age of identification of hearing loss (<6 months, 6 to 12 months, >12 months); and 4) children with hearing technology (cochlear implants, hearing aids).
If appropriate and with sufficient, complete data, we will verify results of the subgroup analyses with univariate meta-regression. The variables outlined above for subgroup analyses will be considered statistically significant at P < 0.01.
Grading the strength of evidence
We will apply methodology developed by the Grades of Recommendation Assessment, Development and Evaluation working group  to rate the evidence for primary and adverse outcomes. The body of evidence for each outcome will be assessed across the domains of risk bias, consistency, directness, precision, and publication bias. The quality of the evidence will be rated as high (very confident that true effect is close to the estimate of the effect), moderate (moderately confident in the effect estimate), low (limited confidence in the effect estimate), or very low (little confidence in the effect estimate). We will discuss the results in light of the strength of findings as well their research and potential practice and family counseling implications.
Reporting of the review findings will follow guidelines for reporting systematic reviews presented in the PRISMA statement . The report will include a flow chart detailing the reasons for included and excluded studies. Guided by knowledge user input, the review findings will subsequently be translated in formats that are adapted to different audiences including clinicians and parents.
Engagement of knowledge users
The principal objective of this review is to understand the effects of using sign language as a support for spoken communication development of children with hearing loss. This project represents collaboration between researchers and three knowledge users representing parents, health and education provider groups with a broad reach, who bring complementary but distinct expertise and strengths that will facilitate knowledge translation. Intervention decisions and parent guidance tend to be based on single studies without attention paid to the quality of the evidence and the particular characteristics of the study population. The expected outcome of partnering closely with end-users is to eliminate some of the potential pitfalls in the way that evidence is interpreted and used. We envision that knowledge translation will take place continually through natural networks and outreach activities, as knowledge users familiarize themselves with the relevant literature and the results of this systematic review. Planned activities include in-service professional meetings, a parent conference presentation, parent information pamphlet, presentation at scientific conferences, peer-reviewed publication, and the integration of the results into academic seminars and coursework.